Study of the Mechanism of Photochemical Carbonylation of Benzene

R. Krishnan and Richard H. Schultz. Inorganic ... Alan S. Goldman and Karen I. Goldberg. 2004,1-43 .... Anne M. Kelly, Glen P. Rosini, and Alan S. Gol...
0 downloads 0 Views 1MB Size
J. Am. Chem. SOC.1994,116, 9498-9505

9498

Study of the Mechanism of Photochemical Carbonylation of Benzene Catalyzed by Rh(PMe&(CO)Cl Glen P. Rosini, William T. Boese, and Alan S. Goldman' Contribution from the Department of Chemistry, Rutgers-The Piscataway, New Jersey 08855-0939

State University of New Jersey,

Received January 19, 1994'

Abstract: Although near-UV irradiation of Rh(PMe3)2(CO)Cl(l) has been shown to lead to efficient carbon-hydrogen bond activation, the title reaction is promoted with much greater efficiency by shorter wavelength (A < ca. 320 nm) irradiation: for example, the carbonylation quantum yield a t 314 nm is found to be 1.5 X lk3vs 2.0 X 10-5at 366 nm. A comprehensive mechanistic study reveals that the shorter wavelength irradiation is required for two separate steps in the catalytic cycle. Photoextrusion of CO, which occurs upon near-UV irradiation and leads to efficient C-H bond activation, is not a major step toward carbonylation under the reaction conditions of the present study nor, apparently, is any other dissociative photoreaction. It is proposed that the first reaction step is the oxidative addition of a benzene C-H bond to an intact photoexcited state of 1. Independently, Field and co-workers have generated and characterized the resulting six-coordinate phenyl hydrido complexes, Rh(PMe3)2(Ph)(CO)HCl(2) under low-temperature conditions. Kinetic studies of the catalytic reaction indicate the involvement of a photoactive intermediate, the decay rate of which is in good agreement with the activation parameters determined by Field et al. for the low-temperature decay of the phenyl hydrido complexes, 2. Short wavelength light is required for the efficient photoreaction of 2, possibly because longer wavelength light is not significantly absorbed, particularly in competition with 1. The carbonylation efficiency increases with C O pressure; this is attributable to the reaction of CO with the unsaturated benzoyl complex, Rh(PMe3)2(COPh)HCl, which results from CO insertion into the Rh-phenyl bond of 2. Reversible loss of benzoyl chloride from the CO addition product, Rh(PMe3)2(COPh)(CO)HCl, is shown to lead to the formation of C&CDO and C&CHO when the reaction is conducted in C&I6/CsD6 mixtures.

Introduction The selective functionalization of hydrocarbons remains one of the most important goals of catalysis.' Carbonylation is one such reaction of particular potential value since aldehydes, particularly linear isomers, are presently produced in large quantities via multistep processes such as cracking, oligomerization, and hydroformylation. Thus it was of considerable interest when, in 1983,Eisenberg reported that Rh(PPh3)2(CO)Cl and several related complexes catalyzed the photochemical carbonylation of benzene.2 Subsequently, Tanaka reported that the trimethylphosphine analogue, Rh(PMe3)2(CO)Cl(l),effected the first example of catalytic alkane carbonylation: photochemical conversion to aldehydes with high selectivity for the terminal position in the case of n - ~ e n t a n e . ~ We have reported that 1-catalyzed alkane photocarbonylation is greatly accelerated by the presence of added aldehyde or ketone and have shown that this system operates via a radical-based p a t h ~ a y .However, ~ in the absence of added organic carbonyl, catalysis by both 1 and Rh(PPh3)2(CO)Cl appears to involve a non-radical mechanism. This is most strongly indicated by the observation that both catalysts exhibit a high selectivity for benzene, a substrate resistant to radical-based pathway^.^,^ (For example, the 1-catalyzed carbonylation of benzene is not accelerated by the presence of aldehyde or ketone4.) In order to Abstract published in Aduance ACS Abstracts, August 15, 1994.

(1) For lead references see: (a) Davies, J. A.; Watson, P. L.;Liebman, J.

F.; Greenberg, A. Selective Hydrocarbon Actiuation; VCH Publishers, Inc.: New York, 1990. (b) Hill, C., Ed. Actiuation and FuncrionalizazionofAIkanes; John Wiley & Sons: New York, 1989. (2) (a) Fisher, B. J.; Eisenberg, R. Organometallics 1983,2,764-767. (b) Kunin,A. J.; Eisenberg, R. J. Am. Chem. SOC.1986,108,535-536. (c) Kunin, A. J.; Eisenberg, R. Organometallics 1988, 7, 2124-2129. (3) (a) Sakakura, T.; Tanaka, M. J. Chem. SOC.,Chem. Commun. 1987, 758-759. (b) Sakakura, T.; Tanaka, M. Chem. Lett. 1987, 249-252. (c) Sakakura, T.; Sasaki, K.; Tokunaga, M.; Wada, K.; Tanaka, M.Chem. Lett. 1988,155-158. (d) Sakakura,T.;Sodeyama, T.;Sasaki, K.; Wada,K.;Tanaka, M. J. Am. Chem. SOC.1990, 112, 7221-7229. (4) Boese, W.T.;Goldman,A. S.J. Am. Chem. SOC.1992,114,350-351.

investigate non-radical-based 1-catalyzed hydrocarbon carbonylation pathways, weinitiated a mechanistic study, choosing benzene as a substrate. (Benzene offers the additional advantage of its inability to undergo dehydrogenation, which can be catalyzed by 1 much more efficiently than carbonylation in the case of alkanes.) Herein we report the results of this study which, in conjunction with an independent study by Field, reveals a catalytic mechanism and relevant chemistry that have several unusual aspects.5 Results and Discussion Irradiation of benzene solutions of Rh(PMe3)2(CO)Cl (1; typically 7-10 mM) under CO atmosphere (0.1-4.0atm) at 50 OC results in the formation of benzaldehyde, eq 1. Other organic

C&,

+ co

1, hv

C&C(O)H

(1)

products are also formed, but initially in much smaller quantities than benzaldehyde. In a typical run (10 mM 1 in benzene; 2.5 h of irradiation, 500 W H g a r c lamp, X > 290 nm, 50 "C, 800 Torr of CO,2 mL) product yields were the following: benzaldehyde, 44.1 mM; benzyl alcohol 1.6 mM; benzophenone, 1 .O mM; biphenyl, 0.5 mM. In an otherwise similar but more extended run (24 h of irradiation, 7 mM l), yields were the following: benzaldehyde, 0.37 M (corresponding to 52 catalytic turnovers); benzyl alcohol 0.19 M; benzophenone, 0.13 M; biphenyl, 0.019 M. Yields of Benzaldehyde. Benzene carbonylation catalyzed by Rh(PPh3)2(CO)Cl upon irradiation was reported by Eisenberg et al. to yield benzaldehyde concentrations equal to the corresponding thermal equilibrium values under several different CO pressures.2 When a benzene solution with a benzaldehyde concentration greater than the equilibrium value was irradiated, (5) Boyd, S. E.; Field, L. D.; Partridge, M. G. J. Am. Chem. SOC. Preceding paper in this issue.

0002-786319411516-9498$04.50/0 0 1994 American Chemical Society

J. Am. Chem. SOC.,Vol. I 16, No. 21, I994 9499

Mechanism of Photochemical Carbonylation of Benzene Table 1. Results from Benzaldehyde Crossover Experiments"

1 2 3 4 5

800 800 100

1600 800

7.0 7.0 7.0 7.0 2.0

34.8 66.2 38.2 40.7 42.0

17.2 0.0 13.0 18.0 15.8

17.5 3.4 16.6 20.3 18.7

30.5 30.4 32.2 21.0 23.5

69.4 59.2 76.6 69.0

Conditions: 50 OC, 1:l C6Ha:C6D6 except entry 2 (neat C6D6 with 4 mM added CsHsCHO). Calculated by taking the sum of the percentages of C6HsCDO and C~HSCHOand doubling to allow for randomly produced C~HSCHO and C&CDO. a

net decarbonylation resulted until equilibrium was achieved. These results strongly indicated that the actual catalytic cycle was nonphotochemical; the role of light was only to generate a thermochemical catalyst, presumably via ligand photoextrusion. In the case of 1-catalyzed carbonylation, however, we find that benzaldehyde concentrations significantly above equilibrium values (ca. 0.002 M/atm of C O at 50 "CZ) can be obtained. For example, in addition to obtaining 0.37 M benzaldehyde under 1 atm as noted above, 0.080 M benzaldehyde was obtained under similar conditions with a CO pressure of only 5OTorr. In addition, 4 mM benzaldehyde added to a CsD6 solution does not show any deuterium incorporation during the reaction (Table l ) , thus suggesting the irreversibility of the elimination of benzaldehyde in the catalytic cycle. These results are not altogether surprising given the ability of 1 to catalyze alkane carbonylation which would yield only undetectable quantities of aldehyde upon reaching thermal equilibrium ( < 1 W M at 298 K),6 in contrast with the significant yields reported by Tanaka.3 Thus, in the case of 1, light drives the catalysis beyond the thermal equilibrium aldehyde concentration for both alkanes and benzene-and therefore at least one of the steps within the catalytic cycle must be photochemical. CO Pressure Dependence. Flash photolysis studies conducted by Ford have revealed that loss of C O is the dominant photoreactionof 1(at wavelengths >330nm) and that the resulting Rh(PMe&Cl fragment oxidatively adds a benzene C-H bond.' C O loss was also found to be the dominant photoreaction of Rh(PPh,),(CO)Cl (A > 315 nm).* In accord with these findings, the rate of benzene carbonylation catalyzed by Rh(PPh3)2(CO)C1 has been found to vary inversely with C O pressures2 Likewise, alkane dehydrogenation catalyzed by 1 is strongly inhibited by CO atmosphere.9 It is thus somewhat surprising that the rate of benzene carbonylation catalyzed by 1 (7.0 mM, 50 OC, 500-W Hg-arc lamp, X > 290 nm) is found to be approximately proportional to the CO pressure at less than 1 atm and appears to approach saturation a t higher pressures (see Figure 1). This finding is difficult (but not impossible) to reconcile with a mechanism in which C O loss is the primary photoprocess. Wavelength Dependence. The quantum yield of reaction 1 was determined at 366 nm to be 2.0 X 10-5 ( I = 3.3 X einstein/s). At 314 nm ( I = 2.8 X 10-8 einstein/s) the quantum yield was found to be 75 times greater, 1.5 X 10-3. This pronounced wavelength effect has significant implications concerning the primary photoprocess of the reaction. The primary quantum yield for C O loss from 1 has been independently determined to be 0.1 at 366 nm.9 Therefore, loss of CO from 1 (6) Calculated according to the following values for AH'r (kcal/mol) and So (cal/mol.deg.K) respectively: hexane(1) (-47.52,70.76); CO(g) (-26.42, 47.04); heptanal(1) (-74.5, 83.3) [obtained from: Stull, D. R.; Westrum, E. F.; Sinke, G. C. The Chemical Thermodynamics of Organic Compounds; Robert E. Kreiger Publishing: Malabar, FL, 19871. (7) (a)Spillet,C.T.;Ford,P. C.J.Am. Chem.Soc.1989,111,1932-1933. (b) Ford, P. C.; Netzel, T. L.;Spillet, C. T.; Porreau, D. B. Pure Appl. Chem. 1990, 62, 1091-1094. (8) (a) Wink, D. A,;Ford, P. C. J. Am. Chem. SOC.1985,107,5566-5567. (b) Wink, D. A.; Ford, P. C. J . Am. Chem. Soc. 1985, 1794-1796. (c) Wink, D. A.; Ford, P. C. J . Am. Chem. SOC.1987, 109, 436-442. (9) (a) Maguire, J. A.; Boese, W. T.; Goldman, A. S. J. Am. Chem. SOC. 1989, 1 1 1 , 7088-7093. (b) Maguire, J. A.; Boese, W. T.; Goldman, M. E.; Goldman, A. S . Coord. Chem. Rev. 1990, 97, 179-192.

= fE

10

?

F r

zc

5

0

0

Figure 1. Rate of benzaldehyde formation as a function of CO pressure.

to generate the 14-electron Rh(PMe3)lCl fragment cannot be the only photochemical process occurring, as this would require an impossibly high quantum yield for C O loss from 1 at 314 nm (7.5, i.e., 75 X 0.1). Significantly, this conclusion is consistent with the above noted (positive) C O pressure dependence of the rate of reaction 1 (Figure 1). These findings leave open three possible roles of irradiation (ca. 314 nm) in reaction 1: (i) the catalytic cycle may involve a single photoprocess, one that is not C O extrusion; (ii) CO photoextrusion may be involved, along with a second photoprocess responsible for themarked wavelength dependence; or (iii) the catalytic cycle may contain two (or more) photoprocesses, neither of which is CO extrusion. Irradiation Intensity Dependence. In an attempt to determine if the catalytic cycle of reaction 1 involved more than one photochemical step, we examined the dependence of the quantum yield on the intensity of irradiation. Using broadband irradiation (7 mM 1,800 Torr of CO, 50 OC, X > 290 nm, 500-W Hg-arc lamp) we found that the rate of benzaldehyde formation was linearly proportional to the light intensity (i.e. the quantum yield was intensity independent) over a 300-fold intensity range (obtained with the useof 200- and 500-W lamps and with neutral density filters). While this result is consistent with a one-photon (single-photoprocess) cycle, it does not preclude a two-photon cycle if the second photochemical step occurs with high efficiency relative to any thermochemical processes of the photoactive intermediate under these reaction conditions. If the second photochemical step of the cycle is occurring with high efficiency under X > 290 nm irradiation, then any increase in the rate of the first photoprocess should be reflected in a proportional increase in the overall rate of benzaldehyde formation. In this context we conducted a "dual-beam" experiment in which the sample was irradiated with two sources simultaneously, a 200-W Hg/Xe-arc lamp with a >290 nm cutoff filter and a much more intense (ca. 5-10 fold) 500-W Hg-arc lamp with a >340 nm cutoff filter. Individually, the 500-W lamp effected a benzaldehyde production rate of 0.068 mM/min as compared with 0.24 mM/min from the 200-W lamp. With the two lamps irradiating simultaneously, the resulting rate was 0.3 1 mM/min, approximately equal to the sum of the single-lamp rates. The 500-W lamp, however, would necessarily greatly increase the extent of C O photoextrusion from 1. We can thus conclude that, even if a two-photon mechanism is operable, the predominant catalytically significant initialphotoreaction is not CO loss. This conclusion is of course consistent with the promoting effect of increased CO pressure noted above (Figure l), as well as the wavelength dependence which, by itself, eliminates the possibility of CO loss as the photoprocess of a one-photon mechanism. The photocatalysis thus originates from an excited state of 1different from the one obtained upon low-energy irradiation. In this context it is probably significant that the above noted wavelength dependence (e.g. 3 14 vs 366 nm) involves irradiation into different

Rosini et ai.

9500 J . Am. Chem. Soc., Vol. 116, No. 21, 1994

o.6i \ 0.4

4

.\

the rate demonstrates that short wavelength light is requiredfor efficient promotion of the second, as well as the first, photochemical step. The requirement of short wavelength light for the irradiation of the photoactive intermediate may simply result from failure of the intermediate to appreciably absorb longer wavelength light, especially in competition with 1. This is especially consistent with the intermediacy of a six-coordinate d6 intermediate (vide infra) since such species are generally bleached in comparison with their d8 four-coordinate precursors." The Primary PhOtQprWeSSIs Not Dissociative. As noted above, both 1-catalyzed alkane photodehydr~genation~ and Rh(PPh3)z(C0)CLcatalyzed benzene photocarbonylation2are inhibited by the presence of even low pressure CO atmospheres. In contrast, the rate of 1-catalyzed benzene photocarbonylation (eq 1; A > 290 nm) is promoted by increased CO pressure under ca. 1 atm, above which it is independent of CO pressure (Figure 1). These results argue against CO photoextrusion as a primary photoprocess, in accord with the conclusion that short wavelength light is required for the primary (as well as the secondary) photoreaction. This conclusion may appear somewhat surprising in view of Ford's finding that CO loss is a major photoprocess of 1 (A > 330 nm) and that the resulting Rh(PMe3)zCl fragment oxidatively adds the benzene C-H bond.7 However, Ford's flash photolysis studies reveal why CO loss is not a viable primary photoprocess for carbonylation in the C O pressure range of the present work. In the flash photolysis studies, a CO pressure of 7.6 Torr was found to result in a 65% decrease in the formation of the benzene C-H addition product, attributable to a very rapid reaction of Rh(PMe3)zCl with C0.7 Extrapolation leads to the conclusion that, under 800 Torr of CO, C-H addition would be inhibited 195-fold. Even if in the absence of added CO Rh(PMe3)zCl undergoes C-H addition with complete efficiency, under 800 Torr of CO the quantum yield for the C-H addition step (314 nm) is predicted to be less than 5 X 1W.I2 Since this value is lower than our value for theoverallquantum yield of benzaldehyde formation (1.5 X under 800 Torr, C O extrusion cannot make a dominant contribution to the catalysis under 1 atm or greater CO pressure. Further confirmation of this conclusion is obtained from experiments using a high-pressure vessel equipped with quartz windows: under 46 atm of CO the rate of benzaldehyde formation (2.87 mM/h) is approximately equal to thatunder 1.Oatm (2.70mM/h),and thereforethecarbonylation quantum yield (314 nm) is approximately 1.5 X le3 even under 46 atm. Under such pressure, however, the quantum yield for C-H addition to the three-coordinate species observed by Ford, extrapolated as above, should be no greater than 1.2 X 10-5.12 Any three-coordinate species resulting from ligand loss from 1-not only Rh(PMe3)zCl-would be expected to readily add C O to give a relatively unreactive four-coordinate complex. Thus, on a less quantitative level, the failure of even 46 atm of C O to inhibit reaction 1 argues not only against CO loss as the primary photoprocess but also against loss of either of the other two ligands as well. This conclusion is further supported by the observation that reaction 1 isuninhibited by the presenceof any of the following

o.2 0.2

-

0.00.0 0.0

0.2 0.2

0.4 0.4

0.6 0.6

0.8 0.8

1.0 1.0

llme between pulses (rec)

Figure 2. Benzaldehyde formationper laser pulse (308 nm excimer laser) vs time delay between pulses.

bands of the electronic absorption spectrum of 1 (A, = 280 and 360 nm, respectively). During the course of this work, Tanaka reported that the rate of reaction 1 , using a 308 nm excimer laser, was not proportional to the frequency of pulses: when a sample was irradiated with 10 pulse/s (Hz), the rate of reaction 1 was reportedly more than ten times greater than that resulting from 1 Hz irradiation.10 We have quantified this experiment (see Figure 2). It is important to note that for a simple one-photon catalytic cycle a horizontal line would be expectedfor the function graphed in Figure 2 (i.e. quantum yield should be independent of the pulsing frequency); the results shown in Figure 2, therefore, argue very strongly against a one-photon mechanism. (We assume that there are two photochemical steps although we cannot rule out the possibility of more than two.) Figure 2 reveals that a plot of the relative quantum yield increases with delay between pulses in a very roughly first-order exponential manner with a decay rate constant of ca. 3.0 s-I. This result is strongly supportive of a catalytic cycle involving a photoactive intermediate, generated by the initial laser pulse, which undergoes thermal decay with a rate constant equal to the exponential term, ca. 3.0 s-l. Thus with increasing delay between pulses, a decreasing concentration of the photoactive intermediate is present. Since the decay process is thermal, these results are complemented by the observation that in experiments using a constant output Hg-arc source (A > 290 nm), a pronounced inverse temperature dependence is found for the rate of eq 1. The rates of benzaldehyde formation in a 7 mM benzene solution of 1 were as follows: 26 OC, 0.58 mM/min; 50 "C, 0.32 mM/min; 77 OC, 0.09 mM/min. Wavelength Effect on the Second Photoprocess. As noted above, using the excimer laser source, a dependence on pulsing frequency (in effect, irradiation intensity) was noted for the rate of reaction 1, in contrast to the intensity-independence found for the broadband irradiation conditions. Thus under the conditions of the laser experiment thermal decay of the photoactive intermediate is significantly competitive with its photoreaction. Using the excimer laser (308 nm, 10 pulse/s), another "dualbeam" experiment was conducted, by simultaneously irradiating the sample with a >360 nm light source (200-W Hg-arc lamp). The combination of short (308 nm) and long wavelength (>360 nm) light resulted in no significant increase in benzaldehyde production above that resulting from laser irradiation alone (7.1 mM/h). This observation further supports the conclusion that the primary photoprocess is not C O photoextrusion which would be accelerated by the added long wavelength light source. Furthermore, since under the conditions of this dual-beam experiment the second photochemical step was not occurring with complete efficiency (as demonstrated by the pulse-frequency dependence), the failure of the long wavelength light to enhance (10) Moriyama, H.; Sakakura, T.; Yabe, A.; Tanaka, M. J. Mol. Coral. 1990,60, L9-Ll2.

(1 1) See for example: (a) Vaska, L.; DiLuzio, J. W. J. Am. Chem. SOC. 1962,84,679-680. (b) Sacco, A.; Ugo, R.; Moles, A. J. Chem. SOC.A 1966, 1670. (12) Since the ratio of CO addition (km) to C-H addition kc^) is 6535 = 1.86 under 7.6 Torr of CO, the ratio under 800 Torr (assuming Henry's law) should be (800/7.6) X 1.86 = 195. Similarly under 46 atm the ratio should be (46(760)/7.6) X 1.86 = 8540. An estimate for the primary quantum yield for CO photoextrusion can be obtained from a determination of the quantum yield (A = 308 nm) for catalyticcyclooctanedehydrogenationobtained intheabsenceofCOatmosphere.whichisO.1 whenX = 308 nm(approximate1y equal to that for X = 366 nm): Maguire, J. M.; Goldman, A. S.Unpublished results. Thisvalue (0.1) is used to calculate thequantum yield for C-H addition to Rh(PMe&CI: under 46 atm, for example, the value would be 0.1/8540 = 1.2 X 1W. Note that even if the quantum efficiency of CO photoextrusion is assumed to be 100%, the calculated quantum yield for C-H addition, 1.2 X l(r, is significantly below the overall carbonylation quantum yield, 1.5 X 1 ~ .

Mechanism of Photochemical Carbonylation of Benzene species in solution: PMe3 (7 mM; higher concentration results in formation of Rh(PMe3)3Cl), Me3PO (20 mM), and CH3CN (0.5 M). In view of the observation that increased CO pressure promotes reaction 1, and that added PMe3 or other ligands do not inhibit the reaction, it would appear unlikely that loss of PMe3 from 1 could be the primary photoprocess. We nevertheless investigated this possibility further by attempting to catalyze reaction 1 using Rh(PMe3)(CO)2C1. CO loss from Rh(PMes)(CO)zCl would give the same intermediate, Rh(PMe3)(CO)Cl, as obtained upon PMe3 loss from 1. Photoextrusion of CO from this dicarbonyl would seem at least as favorable as from 1; even a quantum yield of only ca. 10-4 for CO loss from Rh(PMes)(CO)zCl should then lead to detectable benzaldehyde formation under the conditions of the experiment.13 In fact, Rh(PMe3)(CO)2Cl was found to be completely photoinactive (benzaldehyde formation was below the limit of detection, ca. 0.3 mM), supporting the conclusion that Rh(PMe3)(CO)Cl is not an intermediate in the catalytic cycle. To our knowledge, photochemical halide loss in a nonpolar solvent is an unprecedented process. Further, the chloride-loss product Rh(PMe&(CO)+ should readily react with CO or the added ligands noted above (particularly 0.5 M CH3CN) to give metastable cations that should then recombine with C1- to regenerate 1. To further probe this possibility of a chloride-loss intermediate, however, we conducted reaction 1 in the presence of a source of added C1-, [K(18-crown-6)]Cl- (16 mM). No effect on the rate of reaction 1 was observed. Although it seems unlikely that a radical-based process would show selectivity for benzene over alkane substrates, we considered another dissociative reaction as a primary photoprocess for reaction 1: homolytic cleavage of the Rh-C1 bond. However, the presence of added 9,lO-dihydroanthracene (0.1-0.5 M), a highly effective radical trap, shows no inhibition of reaction 1.14-16 Additionally, a negligible isotope effect was found in experiments using C6H6 and C6D6, kH/kD = 1.2 f 0.2 (both individually and in competition; see Table l), further arguing against any radical mechanism and entirely consistent with a C-H(D) oxidative addition process.17 An Associative Primary Photoprocess: Formation of a Phenyl Hydrido Complex. The above observationsstrongly argue against a primary photoprocess involving ligand dissociation from 1. We therefore propose an associative mechanism. It does not seem plausible that photoinduced CO addition, leading to an 18e complex, could lead to benzene carbonylation. We thus propose that the primary photoprocess of reaction 1 is the photoinduced addition of benzene to an electronically excited state of 1, yielding an 18e phenyl hydrido complex. Such a process is to our knowledge unprecedented; however, independent low-temperature studies by Field et al. during the course of this work are consistent with the same conclusion of photoinduced, associative, benzene (13) Our limit of detection for benzaldehyde formation is ca. 0.3 mM. In order to achieve detectable benzaldehyde levels (after 24 h of irradiation) the rate of benzaldehyde formation would need to be greater than ca. 6 X 10-4 times that resulting from I-catalyzed carbonylation. If I-catalyzed carbonylation proceeded via PMc3 loss, assuming an upper limit of 0.1 for the primary quantum yield for PMe3 loss (a process not detected in Ford's flash photolysis work), then the quantum yield for CO loss from Rh(CO)(PMe,)ZCl required for benzaldehyde detection would be 0.1 X 6 X lo-+. (14) 9.10-Dihydroanthracene would be subject to hydrogen abstraction by free phenyl radicals, yielding benzene and anthracene. The rate constant for this reaction is presumably significantly greater than that for the reaction of phenyl radical with diphenylmethanelJ which can be estimated16 as ca. 1 X IO' M-1s-1. Note that unlike benzaldehyde formation, however, formation of biphenyl is severely inhibited by added 9.10-dihydroanthracene, and thcamount of anthracene formed is equal to the amount of biphenyl formed in the absence of added9,lO-dihydroanthracene.This impliesthat the biphenyl that is formed during the reaction results from a radical pathway. (15) Howard, J. A.; Ingold, K. U. Can. J . Chem. 1968.46, 2661-2665. (16) Fischer, H., Ed. Landolt-Bornstein Numerical Data and Functional Relatiomhips in Science and Technology; Springer-Verlag: Berlin-Heidelberg, 1984; Vol. 13, Subvolume b, pp 26, 28, 213. (17) Jones, W. D.; Feher, F. J. Arc. Chem. Res. 1989,22,9 I and references cited therein.

J . Am. Chem. SOC.,Vol. 116, No. 21, 1994 9501

addition to 1: irradiation (A > 290 nm, -40 " c ) of 1 in a C.&/ T H F (1 :4) solution yields two isomers of Rh(PMe3)2(CO)Cl(Ph)H, 2a and 2b, as determined by 'H, 31P, and 13C N M R spectroscopy. Both isomers have a cis phenyl-hydride arrangement, consistent with oxidative addition to the four-coordinate complex.5

1+w

2a

2b

Significantly, reaction 2 is not significantly inhibited by the presence of 1 atm of CO (0-1 atm).l8 This implies that if formation of 2 were to proceed primarily via CO loss, then the rate of benzene addition to Rh(PMe3)2Cl must be much faster than CO addition, even under 1 atm of CO (a hypothesis inconsistent with the flash photolysis work discussed above).7.* In such a case the reaction rate should be independent of benzene concentration. However, we simultaneously irradiated (A > 330 nm) at -40 "C two T H F solutions of 1, containing 5% and 20% benzene,respectively,under1 atmof CO. After 5 h, the formation of both 2a and 2b was found to be 4-fold greater in the 20% benzene solution (0.42 and 0.26 mM, respectively) than in the 5% solution (0.08 and 0.04 mM).19 Thus the reaction rate is apparently proportional to the benzene concentration; this result in tandem with the CO pressure independence of eq 2 (reproduced under the same conditions as the present benzene-dependence experiment)20rules out the possibility that photoextrusion of CO is a major pathway for reaction 2 under these conditions. Further evidence against the intermediacy of Rh(PMe3)zCl in the short wavelength irradiations was obtained by conducting separate short and long wavelength irradiations of 1 (-78 "C). The resulting distribution of Rh(PMe3)zCl(Ph)(H)(CO) isomers was found to be significantly wavelength dependent: the ratio of 2a:2b was 4.5 vs 1.9 at short (A > 330 nm) and long (A > 360 nm) wavelengths, respectively. This result rules out the intermediacy of ground-state Rh(PMe3)zCl in the short wavelength irradiations, assuming in accord with Ford's work7**that the long Wavelength irradiations proceed via formation of Rh(PMe9)zCl and subsequent addition of benzene to give Rh(PMe&Cl(Ph)(H). Further, this result can be used to exclude any mechanism proceeding via Rh(PMe3)2Cl(Ph)(H). For example, a pathway involving irradiation of 1 to give excited state Rh(PMe3)2Cl, which, hypothetically, would undergo thermal decay or reaction with benzene much faster than CO addition, would afford kinetics identical to those for our proposed mechanism involving an undissociated excited state.2l However, since it shares a common intermediate (Rh(PMe3)2Cl(Ph)(H)) with the long wavelength irradiation pathway, a hypothetical excited state Rh(PMe3)2Cl pathway should give the same distribution of isomers of 2, in contrast with the observed results.22 (18) Field, L.,personal communication. (19) Themerry-go-round typeapparatususedresultedin averylow effective light intensity, resulting in low conversions even after 5 h of irradiation. Irradiating individual samples introduces some comparative error due to variations in focussing the light beam on the 5 mm NMR tube and possible fluctuations in temperature; however, much higher conversions were obtained (leading to the same conclusion of benzene concentration dependence). Specifically, after 2.5 hof irradiation (-40"C)a 5% benzene solution afforded 1.83 mM 2s and 1.03 mM 2b while a 20% benzene solution gave 11.85 mM 2aand 7.67 mM 2b. (20) Simultaneous irradiation of a third sample, with 20% benzene under N2 atmosphere, resulted in the formation of 0.47 mM 2a and 0.37 mM 2b (vs 0.42 and 0.26 mM, respectively, under 1.0 atm of CO). Thus (assuming this difference is not simply due to experimental error) it represents only a 12% increase in the rate of formation of the catalytically relevant complex, 2s. This increase, which likely results from a concomitant pathway involving CO loss and benzene addition to Rh(PMel)zCl, would be expected to be considerably less significant (virtually undetectable) under even low CO pressures (see ref 1l), whichareofcourse necessaryforcatalyticcarbonylation. (21) We thank a referee for suggesting the intriguing possibility of an excited-state Rh(PMe3)zCl intermediate.

9502 J . Am. Chem. SOC.,Vol. 116, No. 21, 1994

Rosini et al.

Bimolecular reactions of organometallic excited state species are rare, and unprecedented in the caseof C-H oxidative additions. However, in retrospect it seems reasonable that the present reaction should be exceptional in this respect for several reasons: (i) Complex 1 is a 16e complex so it is at least formally capable of direct oxidative addition. (ii) Oxidative addition of H2 (electronically related to C-H addition) is well-known for the (ground state) iridium analoguesz3 of 1 and for related trisphosphine-rhodium halide complexes,24and it has recently been shown to occur for 1 itself to give a labile intermediateeZ5(iii) The barrier to oxidative addition to the related complex, transIr(PH3)2(CO)Cl,has been shown by ab initio calculations (in the case of H2) to be largely due to repulsion between the electron pairs of the substrate and the d,z orbitaLZ6 Thus any transition that removes an electron from the dpzorbital (the HOMO of both the Ir complex and l)27 might greatly facilitate addition.28 (iv) The significant kinetic barrier to reductive elimination reactions of octahedral d6complexes29 should serve to stabilize the addition product (which in this case enables both its observation and its subsequent photolysis). In cases where the kinetic barrier is small, even if photoinduced addition can occur, the adduct may be too short-lived to observe if elimination is thermodynamically favorable. Alternatively, if addition is thermodynamically favorable, a small kinetic barrier would enable the corresponding thermal addition to obscure the photoreaction. (v) The occurrence of associative reactions of organometallic excited states is minimized by their generally short lifetimes. In a case such as this however, a reaction with a solvent molecule, the time required to encounter the substrate is obviously very short. (vi) On the basis of the above criteria, a relatively small class of organometallic compounds would appear favorable for photoassociative reactions; of these the best studied are complexes of the form ML3X and ML2(CO)X (where M is Rh or Ir, L is generally a phosphine, and X is a formally anionic ligand). Since steric factors generally play an important role in associative reactions, it is noteworthy that complex 1 is clearly one of the least sterically hindered of this class of compounds. Of the two Rh(PMe&(CO)Cl(Ph)H isomers, only 2a is in the correct geometry to undergo a migratory insertion to form a benzoyl group. Therefore it is 2a that may be considered to be the metastable photoactive intermediate, with the quantum yield for reaction 1 dependent upon a competition between thermal decay of 2a and its photochemical processes. The rate constant for thermal decay of the photoactive intermediate, obtained from the laser pulse experiment described above, is roughly 3.0 s-l (50 "C). This value is, within experimental error, in full agreement with the rate constant for the thermal rearrangement of 2a to 2b (1.7 s-I at 50 "C), independently obtained from low-temperature N M R spectroscopy and extrapolated to 50 0C,5in strong support of 2a as the identity of the photoactive intermediate. Reaction Steps Subsequent to the Formation of 2a. As an octahedral Rh(II1) complex, a photoactive intermediate 2a would (22) Rh(PMe&CI(Ph)(H) would be expected to isomerize quickly. Thus even if C-H addition to excited state and ground state Rh(PMe3)2Cl gave different isomers the resulting CO adducts would be expected to have the same geometry. See, for example: (a) El-Idrissi, I.; Eisenstein, 0.;Jean, Y . New J . Chem. 1990,14,671-677 and references cited therein. (b) Koga, N.; Morokuma, K.J. J . Am. Chem.SOC.1993,115,68836892. (c) Abu-Hasanayn, F.; Krogh-Jespersen, K.; Goldman, A. S . To be submitted for publication. (23) Vaska, L. Acc. Chem. Res. 1968, 1, 335-344. (24) Halpern, J.; Wong, C. S . J. Chem. SOC.,Chem. Commun. 1973,629. (25) Duckett, S. B.; Eisenberg, R.; Goldman, A. S. J . Chem. SOC.,Chem.

Commun. 1993, 1185-1 187. (26) Abu-Hasanayn, F.; Goldman, A. S.; Krogh-Jespersen, K. J . Phys.

Chem. 1993,97, 5890-5896. (27) Brady, R.; Flynn, B. R.; Geoffroy, G. L.; Gray, H. B.; Peone, J.; Vaska, L. Inorg. Chem. 1976, 15, 1485-1488. (28) The band at ca. 280 nm has not been assigned. It is interesting to note, however, that the lowest energy band has been assigned to a d,l bl. tran~ition;~~ since the bl, orbital has largely pz character this might not significantly reduce the barrier to oxidative addition. (29) Collman, J. P.;Hegedus, L. S.; Norton, J. R.; Finke, R.G. Principles and Applications of Organotransition Metal Chemistry; University Science Books: Mill Valley, CA, 1987; pp 324-333.

-.

Scheme 1. Proposed Mechanism of 1-Catalyzed Benzene Photocarbonylation (Short Wavelength Irradiation) PMe,

I

PhCHO LMe,

PMe,

I.

OCRh-CI

I

YH6 c6H7 PMe,

PhCHO

'*

Me,P

L

I

I co

I

,.

I'

3b

Ph

,co

+

CC-Rh-H

CI-RLPh

/I

PMe,

PM~,

2a

2b

A

Scheme 2. Independent Synthesis of Acyl Hydrido Intermediates (and Elimination of Benzaldehyde) Me3P

Me3P

+ co

I

Me3P--Rh

,*GO

Ih o

!OC-+-II-!-P~]

dC'ph 4

7- I

PMe, O C - ~ ~ - ~ ~

/I

8

PMe3

/I

PMe,

8

PhCHO

I

PMe, 1

undoubtedly absorb long wavelength (A > ca. 330 nm) light much less strongly than 1," in accord with the conclusion that the second photoprocess is greatly favored by short wavelengths. Presumably, the catalytically significant photochemical reaction of 2a results in migratory insertion to yield a 16e benzoyl hydrido complex (A in Scheme 1). A subsequent competition between CO addition and migratory deinsertion would explain the observed C O pressure dependence of the rate of reaction 1. Addition of C O to A could produce up to three trans-phosphine isomers of Rh(PMe3)2(CO)(Cl) (C0Ph)H (3). Reductive elimination of benzaldehyde from either 3a or 3b (see Scheme 2) would then complete the catalytic carbonylation cycle (the third possible isomer does not have the correct stereochemistry to reductively eliminate benzaldehyde). The proposed acyl hydrido carbonyl intermediate, 3, was generated in situ. CO (800 Torr) was added to an acetone-& solution of Rh(PMe3)2(CO)Ph (4)30 to give a solution of Rh(PMe3)2(CO)2(COPh). Excess 2,6-di-tert-butylpyridine.HCl(a convenient source of HCI) was then added to the solution at -78 "C, and the reaction was monitored by lH and 3lP N M R spectroscopy (see Scheme 2). The reaction was repeated with W O and monitored by l3C N M R spectroscopy. Only two of the three possible trans-phosphine isomers were observed at -65 OC; (30) Boyd, S . E.; Field, L. D.; Hambley, T I . ; Partridge, M. G. Organometallics 1993, 12, 1720-1724.

Mechanism of Photochemical Carbonylation of Benzene

Scheme 3. Proposed Mechanism of 1-Catalyzed Benzene Photocarbonyl (Short Wavelengths) Including Reversible Loss of ArC(0)Cl To Account for the Results of Crossover Experiments and the Effects of Added Aroyl Chlorides AfC(0)Y

I

OC-Rh-CI

-

ArC(0)Y

PMe3

I I

OC-Rh-Y

PMe3 + ArC(0)CI

A i = C.H,, CIDI, (m-CH,)C,H, Y=H,D

c'-

:kYAi

Me8

0

A

these were spectroscopically determined to be 3a and 3b (Scheme 2). At early reaction times the trans HCl addition product, 3a, was themajor (>70%)product, in accord with theknownaddition of HCl to the related complex 4.5 The concentration of 3a was then seen to slowly diminish with time with the concurrent growth of 3b. At temperatures above -55 OC, 3a completely disappears leaving only 3b, and 1 is observed to begin growing in slowly. 3b decomposes to give 1 and benzaldehyde (quantitative by GC) rapidly at temperatures above -40 OC. This indicates that reductive elimination of benzaldehyde from 3b occurs thermally and very rapidly under the conditions of the catalysis (50 "C), consistent with our assignment of the second photoprocess as being a photoreaction of 2a (and not 3). The rate of thermal decay of 3a and 3b also explains why these intermediates were not observed during Field's experiments which were done at -40 OC, a temperature at which 2a and 2b are stable but 3a and 3b are not. In an attempt to directly investigate the photochemistry of the phenyl hydrido complexes, a sample of 1in THF/C6D6 (4: 1) was irradiated at -40 OC to initially obtain a good conversion to 2a and 2b (some 4 is also formed). The solution was then placed under CO, irradiated at -78 OC, and then monitored by N M R spectroscopy. It was found that small quantities of 3b (but not 3a) were observed under these conditions (see Experimental Section). Formation of Crossover Product. The mechanism indicated in Scheme 1 implies that both the phenyl group and the aldehydic proton of the benzaldehyde product derive from the same molecule of benzene. A series of crossover experiments were conducted in 1:l C ~ H ~ : C ~ and D S ,the products were analyzed by GC-MS (see Table 1). The results indicate that a substantial amount of crossover is occurring, although the amount is less than that which would be expected from a completely random process. A priori, the instability of phenyl radicals would seem to make a radical pathway an unlikely explanation for the observation of crossover product, particularly in view of the selectivity observed for benzene vs alkane carbonylation. Empirically, the lack of inhibition by the presence of a high concentration (0.5 M) of 9,lO-dihydroanthracene (noted above) would seem to rule out the involvement of phenyl radicals. Finally, it should be noted that the fact that the extent of crossover is less than statistical indicates that, regardless of the source of crossover product, a significant non-radical pathway does exist. Field et al. have found that irradiation of 1a t low temperatures yields, in addition to 2a and 2b, small amounts of Rh(PMe&(C0)Ph (4).5 This presumably arises from the loss of HCl from

J. Am. Chem. SOC., Vol. 116, NO.21, 1994 9503 either or both of the phenyl hydride complexes 2 (which is most likely a photochemical step since HCl readily adds thermally to both 1and 4 at the reaction temperature). In the case of C6H6/ C6D6solutions, reversible loss of HCl/DCl from 2 (or 3) would result in the formation of crossover product. To address this possibility, experiments were conducted in which either water (to saturation; 10 pL/mL) or tert-butyl alcohol (0.21 M) was added to the reaction mixture prior to irradiation. If free HCl/DCl were being formed during photolysis, it would be expected to undergo rapid exchange with both the water and the alcohol, converting free DCl to HCl. This would decrease the amount of benzaldehyde formed with deuterium in the aldehydic position. It was found that both water and tert-butyl alcohol had only a marginal effect on the isotopic composition of the benzaldehyde formed, suggesting that most of the 2 and 3 formed do not reversibly lose HCl(DC1) under our reaction conditions. If the benzaldehyde crossover product observed during the photolysis of 1 resulted from a binuclear HCl/DCl scrambling pathway, then the amount of crossover should be dependent on the concentration of 1. However, no such concentration dependence was found (see Table 1). Therefore, these experiments (independence of the crossover on the concentration of 1 or on the presence of added water or terr-butyl alcohol) eliminate the possibility that the observed crossover is due to a pathway that involves reversible loss of HCl/DCl. A series of reactions was carried out in order to further investigate the formation of aldehyde crossover product. It was found that neither added PMe3 nor CO inhibited the degree of crossover, implying that neither of these ligands need to be lost to generate an open coordination site necessary for scrambling to occur. Surprisingly, the degree of crossover was found to increase with increasing CO pressure (Table 1). This can be explained by proposing that the scrambling takes place after coordination of C O to A, and that at lower C O pressures, direct reductive elimination of isotopically pure benzaldehyde (CsH5CHO or C6D5CDO) from A effectively competes with co addition to give 3 which can then lead to isotopic scrambling. Consistent with this interpretation, loss of benzaldehyde from A has previously been postulated by M i l ~ t e i n . ~ ~ Since the loss of neither HCl, PMe3, nor CO was found to be responsible for the majority of the observed crossover, experiments were conducted to determine whether or not the loss of chlorobenzene from 2 or the loss of benzoyl chloride from 3 (in both cases generating the same intermediate, [Rh(PMe&(CO)H]), could be responsible. Added chlorobenzene had no effect on the amount or isotopic distribution of the products. However, addition of (undeuterated) benzoyl chloride to a C6H6/ C6D6 reaction solution resulted in the formation of benzoyld5 chloride and an excess of do and dl aldehyde, indicating that reversible benzoyl chloride loss does occur from 3. Note that loss of benzoyl chloride from Rh(PPh3)2(COPh)(CO)Clz, a complex closely related to 3, has been reported;32 and further, we have found that loss of benzoyl chloride from Rh(PMe3)2(COPh)(CO)C12 occurs and is r e v e r ~ i b l e . ~ ~ In order to further probe the reversible loss of benzoyl chloride from 3, experiments were conducted in which m-toluoyl chloride was added to solutions of 1in C&prior to irradiation. In addition to benzaldehyde-& benzoyl-d5 chloride and m-tolualdehyde-dl were observed. Evidence for the reversibility of the insertion step was also obtained: toluene was formed and the amount decreased with increasing CO pressure (see Experimental Section). This is consistent with Scheme 1 which implies that C O addition to the m-toluoyl analogue of A should prevent deinsertion. Since arene elimination occurs via the aryl hydride complex 2a, irradiation of which leads to CO insertion, Scheme 1 also suggests (31) Milstein, D. Ace. Chem. Res. 1984, 17, 221-226. (32) Stille, J. K.;Regan, M.T. J . Am. Chem. SOC.1974,96.1508-1514.

Seealso: Haynes, A.; Mann, B. E.; Morris,G. E.; Maitlis,P. M.J. Am. Chem.

Soe. 1993, 115, 40934100 and references cited therein. (33) Rosini, G. P.; Goldman, A. S. To be submitted for publication.

Rosini et al.

9504 J. Am. Chem. SOC.,Vol. 116. No. 21, 1994 that the amount of toluene formed should decrease with increasing irradiation intensity. Experiments were conducted using neutral density optical filters, and the ratio of toluene/m-tolualdehyde was monitored by GC. It was found that the amount of toluene was indeed dependent on the intensity (see Experimental Section), supporting the proposal that the insertion step is in fact photochemically induced. PossibleRoleofRh(PMej)z(CO)Ph(4). As notedabove, Field et al. have reported that 4 forms during low-temperature irradiation of 1in benzene/THF (1:4). In addition, these workers have found that 4 is itself a benzene photocarbonylation catalyst. The question thus arises as to how much (if any) of the observed chemistry of 1is occurring through the participation of 4 under our conditions (50 "C). The participation of 4 could readily explain the formation of benzophenone, as well as the observed benzaldehyde crossover, since the aldehydic proton must derive from a second molecule of benzene. Although there is no detectable 4 observed upon monitoring (by IR) a reaction sample run a t 50 "C, the possibilities that small concentrations of 4 strongly absorb short wavelength UV light or that larger concentrations are present during irradiation were considered. An experiment was conducted in order to compare the rates of benzaldehyde formation of 1 and 4. Under identical conditions (10 mM complex, 800 Torr of CO, 50 "C), a solution of 1yielded approximately ten times more benzaldehyde than 4 (44.1 and 4.2 mM, respectively) while yielding somewhat less benzophenone (1 .O and 1.5 mM, respectively). This implies that the majority of the chemistry observed during the photolysis of 1 does not involve the participation of 4. Additionally, even if it is assumed that all of the benzophenone formed during the 1-catalyzed reaction is due to the participation of 4, then an upper limit can be placed on the amount of benzaldehyde formed due to concurrent photochemistry of 4, if it is also assumed that the ratio of benzaldehyde/benzophenone produced by 4 is unaffected by the presence of 1. This leads to the conclusion that, for example, in a typical run where 44 mM benzaldehyde and 1.0 mM benzophenone are formed during the photolysis of 1, only 3 mM benzaldehyde may be due to concurrent photochemistry of 4. (Of course it is entirely possible that some or all of the benzophenone does not arise via a 4-catalyzed pathway; 3 mM is therefore an upper limit.) Summary and Conclusions. A mechanism of the full catalytic cycle of the 1-catalyzed photochemical (short wavelength) carbonylation of benzene is proposed. The most unusual aspect of the cycle is that the primary photoprocess does not involve ligand loss. Photochemical CO loss from 1 is known to lead to C-H addition' and, in the case of alkanes, catalytic dehydrogenation.9 Similarly, loss of a ligand (either PPh3 or CO) from Rh(PPh3)2(CO)Clleads to benzene carbonylation.2 In the present case, however, the C-H addition step proceeds predominantly via an unrelated, associative, process. This conclusion is based on studies of the dependence of the catalytic and stoichiometric reaction rates on C O pressure, wavelength, and irradiation intensity, the results of which are anomalous when compared with the behavior of these other systems which operate via addition to three-coordinate rhodium(1). Note that in the case of 1-catalyzed alkane carbonylation, the most effective conditions involve an entirely different mode of breaking the C-H bond (hydrogen abstraction by a photoexcited organic carbonyl), independent of any rhodium complex.4 In spite of the different modes of C-H bond activation, the 1-catalyzed alkane dehydrogenation and the Rh(PPh3)2(CO)C1-catalyzedbenzene carbonylation systems share with the present system important mechanistic features characteristic of organorhodium-catalyzed chemistry. All threesystems (and the radicalbased alkane carbonylation system as well) depend upon the ability of rhodium to shuttle between the I and I11 oxidation states, and all involve both six- and five-coordinate Rh(II1) aryl, alkyl, and/ or benzoyl hydrido intermediates. In all three cases the ability of the 16e square planar precursor to remain unsaturated, even

in the presence of CO or HZ(conditions which would lead to 18e complexes in the case of iridium analogues), is critical for the catalysis. The small steric bulk of the PMe3 ligand (in contrast, most notably, with PPh3) favors the stability of the crowded Rh(111) intermediates; this also applies to the radical-based system4 and presumably to carbonylation catalyzed by Rh(PMe3)2(CO)Ph.5 The combination of all these features, in addition to the well-established ability of Rh-C u bonds to undergo insertion, should perhaps be viewed as the key to the hydrocarbonfunctionalization properties of 1 and related complexes, rather than any particular ability to react with C-H bonds.

Experimental Section GeneralProcedures. All samples were handled either under a nitrogen atmosphere in a Vacuum Atmospheres Dry-Lab glovebox or under vacuum. Deuterated solvents were distilled under vacuum from NaK alloy (except for acetone which was dried with molecular sieves). All other solvents were distilled from dark purple solutions of benzophenone ketyl and stored in the glovebox. Benzene was purified prior to drying by a standard procedure to remove thiophene and alkene im~urities.3~ NMR spectra were recorded on a Varian XL-400 NMRspectrometer. Concentrations of organic products were quantified by gas chromatography using a Varian 3400 gas chromatograph with a flame ionization detector and a 50-m methyl silicone gum capillary column. GC-MS was performed using either a Hewlett Packard 5890 Series I1 gas chromatograph/597 1 mass spectrometeror a Finnigan-Mat 8230 high resolution magnetic sector mass spectrometer. The irradiationsemploying constant output sources were carried out in an apparatus described previously? The laser experiments employed a Lambda-Physikpulsed excimer laser charged with XeCl for an output wavelength of 308 nm. Pulsing frequencies were determined with an oscilloscope and the power was determined with a joulemeter. The power remained relatively constant during the course of the experiments at 50 mJ/pulse. Preparation of Rh(PMe3)2(CO)Cl (1). [Rh(COD)Cl]z (2 g) was dissolved in 75 mL of toluene and stirred in an ice bath. Carbon monoxide was then bubbled through the stirring solution for 30 min. PMe3 (1 -66 mL) was then added dropwise over a 5 min period, at which time the solution turns bright yellow. The solution is then allowed to warm to room temperature over 30 min. The solvent was then removed in vacuo, and the product was recrystallized from hexanes (2.06 g, 80% yield). Preparation of Rb(PMeo)z(CO)Ph (4). Rh(PMes)z(CO)Me30 (80 mg) was dissolved in a mixture of 2 mL of benzene and 8 mL of THF. The solution was then irradiated at 0 "C for 2 h. The solvent was then removed in vacuo, and the crude product was sublimed at 70 "C to give a yellow powder (75 mg, 78% yield). Typical Conditions for 1-Catalyzed Carbonylation Reactions. In general, 2-mL samples of 7-10 mM 1 in benzene were placed under CO pressurein an optical glass cuvette (A > 290 nm cutoff) sealed to a ballast used to maintain constant partial gas pressures and equipped with ports for attachment to a vacuum/Schlenk line and for removal of microliter samples for GC analysis. Samples were typically irradiated for 2.5 h in a 50 "C water bath with a 500-W Hg-arc lamp while being stirred magnetically. The samples were then analyzed by GC or GC/MS for the determination of the products, calibrated in all cases with the use of commercially obtained authentic samples. Effect of CO Pressure on the Rate of Aldehyde Formation. A stock solution of 2 mM 1in benzene was prepared. Samples (2 mL) were then irradiated at 50 "C under varying pressuresof CO, and the products were analyzed by GC. The rate of benzaldehyde formation for each run was determined and is shown in Figure 1. Quantum Yields at 314 and 366 nm. Two 2-mL samples of 2 mM 1 in benzene were placed under 800 Torr of CO and irradiated using either a 314 nm interference filter or a combination of 0-52 and 7-60 filters for 366 nm. Actinometry was done using a 2-mL solution of Aberchrome 54035 in toluene in a 1 cm quartz cell to determinethe intensity (einstein/ s) of the light at each wavelength. The amount of benzaldehyde formed was then determined by GC for each experiment,and the quantum yield was determined to be 1.5 X lo-' at 314 nm and 2.0 X le5 at 366 nm. Intensity Dependence Studies. A stock solution of 7 mM 1 in benzene was prepared, and 2-mL samples under 800 Torr of CO were irradiated at 50 "C. The intensity of the light was changed by using neutral density (34) Perrin, D. D.; Armarego, W. L. F.;Perrin, D. R. In Purijcation of Laboratory Chemicals; Pergamon Press: Oxford, 1980; pp 118-1 19. (35) Heller, H. G.; Langan, J. R. J . Chem. SOC.,Perkins Trans. 1 1981, 341.

J. Am. Chem. Soc., Vol. 116, No. 21, 1994 9505

Mechanism of Photochemical Carbonylation of Benzene Alters and by varying the source using either a 500-W or a 200-W lamp. The rateof benzaldehyde formation was then determined for each sample. Dual-BeamExperiments. Two milliliters of a 7 mM benzene solution of 1 was placed under 800 Torr of CO and irradiated at 50 OC with two sources, both individually and simultaneously (at 90° to each other). In the first experiment, the sources were a 200-W lamp (>290 nm) and a 500-W lamp (>340 nm), and in the second experiment the sources were a 200-W lamp (>340 nm) and a 308 nm excimer laser (10 pulse/s). The rate of aldehyde formation was determined for each experiment. As in all photochemical experiments, light sources were carefully focussed to irradiate a large area (ca. 0.3-0.8 cm in diameter) centered on the midpoint of the irradiated windows. Note also that the apparently long lifetime of the photoactiveintermediate (ca. 0.3 s asobtainedfromsingleexcimersource experiments) would allow thevigorous magnetic stirring (ca. 1000 rpm) to ensure a homogeneous distribution of the intermediate. Effect of Laser Pulse Rate. Samples (2.5 mL) of 0.2 mM 1in benzene were placed under 800 Torr of CO and irradiated with a 308 nm excimer laser. The pulsing frequency was varied from 25 to 0.5 Hz, and the rate of aldehyde formation and quantum yield was determined for each experiment by GC. The graph in Figure 2 summarizes the data. Effect of Added Ligands on Photolysis of 1. To each of three samples of 2 mL of 7 mM 1 in benzene was added 7 mM PMep,20 mM MepPO, 0.5 M CHoCN, and 16 mM [K+18-crown-6]Cl. Each sample was then placed under 800 Torr of CO and then irradiated at 50 OC. The rate of benzaldehyde formation was then determined for each sample and compared to a control sample that had no extra ligand added to it. It was found that the rate was unaffected in each of the experiments performed. AttemptedCarbonylationwithRh(PMe~)(C0)2CI.Rh(PMe,)(C0)2CI was generated by adding CO atmosphere to a solution of [Rh(PMeo)(CO)C1]2 (2 mg/2 mL; vco = 1975 cm-l) resulting in a rapid color change and the conversion to Rh(PMe,)(C0)2Cl as revealed in the IR spectrum (YCO = 2001,2086 cm-I), as has been described previously.36 The solution was then irradiated at 50 OC for 69 h, and GC analysis showed that there was less than 0.3 mM benzaldehyde formed.12 WavelengthDependenceof the Ratio of 2a:Zb. A 10 mM stock solution of 1 (12.4 mg) in C6H6 (0.8 mL)/THF (2.75 mL)/acetone-& (0.35 mL) was prepared. Samples in NMR tubes were subjected at -78 OC (dry ice/acetone) to irradiation of either X > 330 nm (1 cm acetone, which transmits the 334 nm Hg-arc line) or X > 360 nm (two Corning 0-52 filters). 3lP NMR spectra were obtained at -80 OC. The ratio of 2a:Zb was determined on the basis of integration of the peaks at 6 -6.5 and -8.5, respectively,and wasperiodicallymonitored. The ratiovariedonly slightly over time; values for the X > 330 nm experiment were the following (time, ratio): 10 min, 4.66; 15 min, 4.74; 20 min, 4.79; 40 min, 5.13; 60 min, 5.40. Values for the X > 360 nm experiment were the following (time, ratio): 20 min, 2.02; 30 min, 2.08; 40 min, 2.12; 50 min, 2.19; 60 min, 2.23. The values of 4.5 and 1.9 were obtained by extrapolation to time = 0 to discount the effect of any secondary reactions of 2a or 2b (although such reactions would in any event be minimal due to the thermal stability of Z5 and its presumably weak absorbance at X > 330 n m l ) . Synthesis of Rh(PMe3)2(CO)(COPh)(CI)H (3). A sample of 4 in acetone-ds was placed under 800 Torr of CO in a resealable NMR tube. This generated only Rh(PMe3)2(CO)2(COPh) in solution by ,lP NMR at -65 OC (the spectrum was broad at room temperature, suggesting reversible CO loss). The CO was then removed in vacuo at -78 "C and then an excess of 2,6-di-ferf-butylpyridine-HCl was added, followed by a freezepumpthaw cycle. The reaction was then monitored by NMR spectroscopy at -65 OC, and two new complexes were seen to grow in. The reaction was repeated with 13C0, and this in conjunction with the previous experiment allowed the two isomers to be unequivocally assigned as 3s and 3b. 39: IH NMR (acetone-&) 8.938 (m, IH), 8.317 (m, 2H), 8.093 (m, 2H), 1.683 (br pseudo-triplet, PMe3), -13.406 (dt, J H - =~ 28.5 Hz, JH-P= 12.2 Hz). 31P NMR (acetone-&): -9.05 (dd, JP-H= 12.0 Hz, Jp-Rh = 99.7 Hz); 13CNMR (acetone-&) 241.0 (m, C(O)Ph), 188.7 (m, CO). 3b: 'H NMR (acetone-&) 8.906 (m, lH), 8.306 (m, 2H), 8.145 (m, 2H), 1.789 (br pseudo-triplet, PMe3). -7.544 (dt, JH-U = JH-P = 12.9 Hz ( J H ~61 Hz)). 31PNMR (acetone-ds) -8.40 (dd, JP-H= 12.2 Hz, Jp-Rh = 105.4 Hz). 13CNMR (acetone-ds) 241.9 (dt, Jpx 7.5 Hz, J R h x = 33.0 Hz, J H < ~5 Hz, C(O)Ph), 185.6 (m, CO, Jc-H 62.1 HZ, JC-Rh = 44.2 HZ,CO). Direct Observation of 3 during Photolvsis of 1. A 50 mM sample of 1 in THF/benzene (4:l) was irradiated ai -40 OC for 2.5 h giving 65% (36) Abu-Hasanayn, F.; Goldman, M. E.; Goldman, A. S.J . Am. G e m . SOC.1992, 114, 2520-2524.

a

conversion of 1 to a mixture of products (5:2a:2b:4 = 13:25:13:12). A second sample was irradiated for 5 h giving a 99% conversion (5:Za:Zb:4 = 12:53:21:13).37 Thesesamples werethenplacedunder 800TorrofCO and irradiated at -78 OC in an attempt to detect the formation of 3. In both samples a small amount of 3b was observed to grow in (in both cases the amount of 3b:Za was roughly 6.5%). Crossover Studies. A 7 mM stock solution of 1 in a 1:l CsD6:CsHs solvent mixture was prepared and 2-mL samples were irradiated at 50 OC under the conditions stated below. The isomeric distribution of the benzaldehyde formed was determined by GC/MS. The crossover experiments with varying C O pressure are summarized in Table 1. Two samples were prepared with either 0.3 pL of PMep or nothing added (control) to a 1-mL sample of the stock solution prior to irradiation, and the samples were then placed under 800 Torr of CO and irradiated for 2.5 h at 50 OC. The PMep was found to have had no effect on the total percentage of crossover. Two samples were prepared adding either 10 pL of water or 20 pL of zert-butyl alcohol (2.5 h irradiation at 50 "C), and the amount of CsHsCDO formed was compared to a control sample in which nothing was added. The results showedminimalchange between thecontrolsampleand theadded waterandalcohol (17.1Sversus 16.0% and 15.3%, respectively). Effect of Added Chlorobenzene. To a 1-mL sample of 10 mM 1 in C6D6 was added 2 pL of chlorobenzene. The sample was then placed under 800 Torr of C O and irradiated for 2.5 h at 50 OC. Analysis of the sample by GC/MS showed that there was no chlorobenzene-d5 or benzaldehyde-dl formed, indicating that reversible chlorobenzene loss is not occurring. Effect of Added Benzoyl Chloride. To a I-mL sample of 10 mM 1 in C6H6/C& was added 0.3 pL of benzoyl chloride. The sample was then placed under 800 Torr of C O and irradiated for 2.5 h at 50 OC. Analysis of the sample by GC/MS showed that there was 30% benzoyl-d5 chloride formed along with an excess of benzaldehyde-do,dl (do,dl:ds,d6 = 68:32 compared to 5050 for the control experiment), indicating that reversible benzoyl chloride loss is in fact occurring. Effect of CO Pressure on the mToluoyl Chloride Reaction. Two 2-mL samples of 10 mM 1 in C6D6 were prepared, to which were added 2.65 pL of m-toluoyl chloride. One sample was then placed under 800 Torr of CO, while the other sample was placed under 50 Torr of CO (the balance being Ar). Each sample was then irradiated for 5 h at 50 OC, and then the m-tolua1dehyde:toluene ratio was determined by GC. It was found that under 800 Torr of CO the ratio was 2.4, while under SO Torr of CO the ratio was only O S , thus indicating that toluene formation is inhibited by increasing CO pressure. Effect of Light Intensity on the mToluoyl Chloride Reaction. Two 1.5-mL samples of 7 mM 1 in C6D6 were prepared, to which were added 2.0 pL of m-toluoyl chloride. Each sample was then placed under 50 Torr of CO (the balance being Ar) and then irradiated for 5 h at 50 OC. One sample was irradiated at full intensity, while the other sample had an 0.5 optical density filter. The m-tolua1dehyde:toluene ratio of each sample was then determined by GC. It was found that the ratio for the full intensity sample was 3.7, while the ratio for the 0.5 OD sample was 2.5, thus indicating that aldehyde formation is enhanced by increasing light intensity. Carbonylation with 4. One milliliter of a 10 mM solution of 4 in benzene was placed under 800 Torr of CO and irradiated at 50 OC for 2.5 h. A second sample of 1 mL of a 10 mM solution of 1 in benzene was also irradiated under identical conditions. It was found that 1 catalyzed carbonylation roughly 10 times faster than 4 under identical conditions. Products for 4: 4.2 mM benzaldehyde, O.OmM benzyl alcohol, 0.2 mM biphenyl, 1.5 mM benzophenone. Products for 1: 44.1 mM benzaldehyde, 1.6 mM benzyl alcohol, 0.5 mM biphenyl, 1.0 mM benzophenone.

Acknowledgment. Support for this research by National Science Foundation G r a n t C H E - 9 121695 is gratefully acknowledged. A.S.G. thanks the Camille and H e n r y Dreyfus Foundation for a Teacher Scholar Award a n d the Alfred P. Sloan Foundation for a Research Fellowship. Prof. L. D. Field is thanked for extensive discussions a n d release of results prior to publica(37) Product 5, which hasspectroscopic properties similar to2a, is tentatively assigned as the analogous THF addition product, Rh(PMel)z(CO)ClH(tetrahydro-2-furyl).